Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
cartpole: struct<random: double, ref: double>
  child 0, random: double
  child 1, ref: double
mountaincar: struct<random: double, ref: double>
  child 0, random: double
  child 1, ref: double
catch: struct<random: double, ref: double>
  child 0, random: double
  child 1, ref: double
pong: struct<random: double, ref: double>
  child 0, random: double
  child 1, ref: double
pendulum: struct<random: double, ref: double>
  child 0, random: double
  child 1, ref: double
parity8: struct<random: double, ref: double>
  child 0, random: double
  child 1, ref: double
add3: struct<random: double, ref: double>
  child 0, random: double
  child 1, ref: double
cmp3: struct<random: double, ref: double>
  child 0, random: double
  child 1, ref: double
to
{'cartpole': {'random': Value('float64'), 'ref': Value('float64')}, 'mountaincar': {'random': Value('float64'), 'ref': Value('float64')}, 'catch': {'random': Value('float64'), 'ref': Value('float64')}}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 149, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 129, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 544, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 3052, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 558, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 361, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2381, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              cartpole: struct<random: double, ref: double>
                child 0, random: double
                child 1, ref: double
              mountaincar: struct<random: double, ref: double>
                child 0, random: double
                child 1, ref: double
              catch: struct<random: double, ref: double>
                child 0, random: double
                child 1, ref: double
              pong: struct<random: double, ref: double>
                child 0, random: double
                child 1, ref: double
              pendulum: struct<random: double, ref: double>
                child 0, random: double
                child 1, ref: double
              parity8: struct<random: double, ref: double>
                child 0, random: double
                child 1, ref: double
              add3: struct<random: double, ref: double>
                child 0, random: double
                child 1, ref: double
              cmp3: struct<random: double, ref: double>
                child 0, random: double
                child 1, ref: double
              to
              {'cartpole': {'random': Value('float64'), 'ref': Value('float64')}, 'mountaincar': {'random': Value('float64'), 'ref': Value('float64')}, 'catch': {'random': Value('float64'), 'ref': Value('float64')}}
              because column names don't match

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Neurosurgeon-Bench: result files

Result data for Neurosurgeon-Bench: can an LLM write and edit neural-network weights directly, with no gradients and no training?

No model weights are released here. This is a benchmark-results dataset (jsonl rows with one record per model and run); the benchmark code, harness, explainer video and README are in the GitHub repo.

Files

file content
results/write_r1.jsonl Track W round 1: 3 games (CartPole, MountainCar, Catch), one-shot hand-written MLP weights, one row per run; rows where the API call failed are removed
results/pilot_write_v2.jsonl Track W v2 pilot: 8 tasks (adds Pong, Pendulum, parity8, add3, cmp3), 6 models
results/pilot_edit.jsonl Track E pilot: repair a policy with 6 corrupted weights using at most 10 edits, 6 models
results/readme_data.json derived numbers used in the README (computed by tools/readme_data.py)
configs/ random-floor and reference baselines, model list

Scoring

Index = mean 100 * q / (1 + q), q = max(0, score - random) / (ref - random). Matching the reference (a CEM-trained policy, not PPO) scores 50; random scores 0. Track E recovery is (edited - corrupted) / (clean - corrupted) clipped to [0, 1]. Index values are therefore not comparable with the original PPO-based Writing Weights Index.

Headline numbers (from results/readme_data.json)

  • Track W round 1: 71 models scored, best index 52.2, median 47.3, 28 models at 50 or more, 18 at 0.
  • v2 pilot: 6 models, best index 54.4. Track E pilot: 6 models, best index 50.2.
  • Track S (neuron steering) has code in the repo but no result files yet.

Pilots are small (6 models, one sample each). An index of 0 means no better-than-random weights were scored.

Load

import json
rows = [json.loads(l) for l in open("results/write_r1.jsonl") if l.strip()]
best = max(rows, key=lambda r: r["INDEX"])
Downloads last month
36